Basel III did not eliminate risk inside the largest bank holding companies. It relocated it. The capital that made bank subsidiaries demonstrably safer ...
Basel III did not eliminate risk inside the largest bank holding companies. It relocated it. The capital that made bank subsidiaries demonstrably safer after 2015 was not raised in the market; it was drained from the nonbank affiliates sitting under the same corporate umbrella. The regulated bank got its buffer. The holding company as a whole did not necessarily get safer, and the part that got thinner is the part supervisors examine least. That is the read the research from Cetorelli and Kundu forces, and it is one the standard "banks are better capitalized than ever" narrative skips entirely.
The Capital Did Not Come From Outside
The intuitive story of post-crisis capital regulation is that banks confronted higher requirements and answered by raising equity. Some did. But the mechanism the authors document is different and more interesting: bank holding companies met Basel III largely by reshuffling equity they already owned, upstreaming it from nonbank subsidiaries into the regulated bank. No new external capital was required to hit the ratio. The consolidated entity simply moved money from a pocket the regulator does not risk-weight into the pocket it does.
On the bank's own scorecard, this is a clean success. Capital ratios rise. Charge-offs fall. Asset quality improves. The classic fear that higher requirements push banks to rebuild return on equity by taking on more risk does not materialize; the opposite happens. Banks cut risk-weighted assets relative to total assets and tilt toward securities and cash. Leverage is essentially flat. The marginal injected dollar of equity shows no juiced return. Judged at the level of the bank alone, the regulation did exactly what it was designed to do.
The problem is that "at the level of the bank alone" is the wrong unit of analysis for financial stability. Distress does not respect legal-entity boundaries inside a holding company. And the balancing entries to the bank's improvement landed somewhere.
Where the Balancing Entry Landed
The nonbank affiliates that funded the banks' capital build moved in the opposite direction, and the timing is not ambiguous. Event-study estimates in the underlying Staff Report show sharp breaks beginning in 2015:Q1, the exact quarter Basel III became binding. Nonbank equity-to-asset ratios fall. Dividends paid up to the parent rise. Net balances owed to parents and affiliated entities increase. This is the transfer, visible in the accounts.
But the accounting entries understate what happened, because the nonbanks did not simply hand over equity and carry on with the same business. As equity became scarcer, they changed what they do. They retreated from equity-intensive lines, trading, advisory, venture capital, and expanded into leveraged lending, particularly consumer credit. The organizational chart itself adapted: holding companies added nonbank lending subsidiaries and shed capital-heavy insurance affiliates. Credit intermediation migrated within the firm, from the regulated bank toward the affiliates that carry lighter capital and lighter supervision.
Read together, these are three compounding shifts, not one. The nonbank now holds a thinner equity buffer. It runs a riskier, more leveraged book. And it is more financially entangled with the parent through the intercompany balances it now owes. Each on its own is manageable. Stacked, they describe an entity with less capacity to absorb a shock and more capacity to transmit one back to the organization that owns it.
Why This Is a Transmission Problem, Not an Accounting One
The reason this matters for stability rather than bookkeeping is the direction of contagion. Internal capital markets are usually described as a strength: a diversified firm can move capital to where it is needed. That is true when the affiliate is a source of strength. It inverts when the affiliate is the weak point. A nonbank with a thin buffer, a leveraged consumer-credit book, and large net obligations to affiliates is precisely the entity that, under stress, forces the parent to choose between supporting it and letting distress propagate.
The bank subsidiary looks pristine on examination. The consolidated organization carries a leveraged, lightly-capitalized lending operation whose troubles would land back on the same balance sheet the regulator certified as safe. The safety is real at the sub-entity level and partly illusory at the level that actually determines whether the firm survives a downturn. That gap between the regulated unit's health and the consolidated firm's health is the hidden fragility in the title, and it is a designed-in feature of measuring capital adequacy one legal entity at a time.
The Case Against Reading This as Fragility
The honest counterargument is that this may be capital allocation working as intended rather than risk-hiding. Equity is expensive; parking it in a low-return insurance affiliate when it is needed to satisfy a binding bank constraint is efficient, not reckless. If nonbank affiliates were overcapitalized relative to their actual risk before 2015, then drawing them down toward a sensible level is prudent housekeeping, and the shift toward consumer lending simply reflects where the firm now sees returns. On this reading, the consolidated entity did not get riskier; it got less wasteful.
That case has real force, and it is the fact that would break the fragility thesis. What it does not resolve is whether the resulting nonbank buffers are thin relative to the new riskier book, not the old one. The affiliates did not just shed excess equity and stand still; they simultaneously levered up and moved into a more cyclical lending mix. A buffer that was comfortable against an advisory-and-trading business is not obviously comfortable against a leveraged consumer-credit book. The efficiency reading requires that the equity drawdown and the risk-up move roughly offset. The evidence that they did offset is not in hand. Until it is, "risk moved to a less visible part of the firm" remains the cleaner interpretation than "risk was optimized away."
What Would Confirm the Read
The observable condition is a credit downturn that stresses consumer lending. If the migrated risk is real, the affiliates that absorbed it will show it first, thinner buffers eroding against rising past-due and nonaccrual balances, followed by pressure on the parent to inject support back into entities it had been drawing from. The tell will be intercompany flows reversing direction: capital that moved down to the bank in 2015 moving back out to prop up the nonbank when its book sours.
For anyone pricing the large diversified holding companies, the practical implication is that bank-level capital ratios are a less complete measure of the organization's resilience than they were before Basel III, precisely because the regulation incentivized moving risk to where those ratios do not look. The consolidated leverage and the composition and capitalization of the nonbank affiliates carry information the headline capital ratio no longer contains. The next stress test worth running is not the one on the bank. It is the one on the affiliate the bank now leans on.
